Mms19 promotes spindle microtubule assembly in neural stem cells through two distinct pathways

Mitotic divisions depend on the timely assembly and proper orientation of the mitotic spindle. Malfunctioning of these processes can considerably delay mitosis, thereby compromising tissue growth and homeostasis, and leading to chromosomal instability. Here we identified Mms19 as an important player in these processes as it promotes spindle and astral microtubule (MT) growth and consequently regulates spindle orientation and mitosis duration in Drosophila neural stem cells. Loss of functional Mms19 drastically affects the growth and development of mitotic tissues in Drosophila larvae. We found that Mms19 performs its mitotic activities through two different pathways. By stimulating the mitotic kinase cascade, it triggers the localization of the MT regulatory complex TACC/Msps (Transforming Acidic Coiled Coil/Minispindles, the homolog of human ch-TOG) to the centrosome. In addition, we present evidence that Mms19 stimulates MT stability and bundling by binding directly to MTs.

Running title: 34 Spindle assembly by Mms19 35 36 Key words: 37 mitotic spindle, microtubule, Mms19, TACC, mitotic kinases, CAK complex 38 39 Introduction the identical strategy. This was not the case, as both CBs and OLs remained smaller (Fig 1A-148 G). In da>CAK, Mms19 P brains, the number of CB NBs was similar as in the wild type (Fig 1D), 149 even though the overall brain volume was reduced. These observations indicated that the 150 Mms19 P CB NBs probably did not proliferate enough to produce the normal amount of 151 neuronal tissue, and that this defect might not only be caused by insufficient CAK activity. 152 This result therefore points to the possibility that Mms19 acts through two different pathways 153 to achieve normal organ size. results are represented as a percentage of the total number of cells per brain lobe. We 164 observed that about half the cells were not labelled (i.e. G1/G0 cells: 47-56%; Fig S1B) and 165 the relative differences between the genotypes were small. However, lack of Mms19 caused 166 a clear increase in the fraction of cells in M phase (35% compared to 24% in the wild type; Fig  167   S1E). This result could either mean that more cells undergo divisions or that the Mms19 P NBs 168 are either trapped in M phase or proceed more slowly through it. 169 Interestingly, the small class of cells that are EdU + and pH3 + showed the highest relative 170 increase in frequency when CAK was overexpressed (Fig S1D; 6.5% vs 3.2% for Mms19 P ). 171 Double positive cells have gone through S phase and were in M-phase at the time of fixation. 172 The fact that overexpression of CAK leads to a higher frequency of double positive cells is 173 consistent with the idea that these cells are progressing through the cell cycle more rapidly 174 and/or that they enter M-phase prematurely. 175 176 NBs depend on Mms19 for timely and coordinated spindle assembly and spindle 177 orientation 178 To study how Mms19 contributes to spindle assembly and progression through mitosis, we 179 utilized a transgene that expresses EB1::GFP, a MT plus end binding and tracking protein that 180 labels growing MT ends (Zhu et al, 2009). Live imaging of NBs expressing EB1::GFP revealed 181 the dynamics of the formation of the mitotic spindle and allowed us to estimate the duration 182 of mitosis, which we defined as the period from the onset of the Nuclear Envelope Break-183 Down (NEBD) until the end of cytokinesis. NEBD onset was determined by the appearance of 184 the GFP signal in the nuclear region, which lacks a GFP signal until NEBD. NBs expressing 185 EB1::GFP in the wild-type background finished cytokinesis approximately 10 min after NEBD 186 (Fig 2A, mov 01). On the other hand, NBs expressing EB1::GFP in an Mms19 P background 187 reached cytokinesis only around 20 min after NEBD (Fig 2B, mov 02). 188 Interestingly, in some of the EB1::GFP expressing Mms19 P NBs, the spindle formation started 189 before the two centrosomes had finished migrating to the opposite sides of the nucleus (Fig 190 2B, mov 03). As a result, at 3 min, a kinked spindle was observed, which eventually 191 straightened out at 5 mins. Around 10% kinked spindles were found in Mms19 P NBs. In one 192 case of a Mms19 P NB spindle, only one centrosome started nucleating MTs (Fig 2C). The 193 spindle remained monopolar until 6 min post NEBD and only then, a bipolar spindle became 194 apparent. Further, around 20% of the spindles in Mms19 P NBs changed their orientation 195 during the course of mitosis ( Fig 2B, B', mov 03), while all wild type spindles examined 196 remained firmly anchored at the cortex and did not change their orientation (Fig 2A, A'). 197 Drosophila larval NBs are characterized by a defined apical-basal polarity (Homem et al, 2012) 198 where the atypical Protein Kinase C (aPKC)-Bazooka-Pins complex localizes to the apical 199 cortex while Miranda localizes to the basal cortex ( Fig S4A). As differentiation of the GMC 200 relies on the inheritance of the Mira-bound cargo, the orientation of the mitotic spindle is 201 tightly coupled to this apical basal polarity axis (Schuldt et al, 1998;Rolls et al, 2003). We 202 therefore further examined the spindle orientation defects in fixed cells labelled with the 203 polarity marker Miranda (Mira), which localizes to the basal cortex of the NB, forming a 204 crescent-like pattern. We found that most of the WT spindles form at an angle of 10 degrees 205 or lower relative to Mira. On the other hand, Mms19 P NBs more frequently failed to align 206 their spindles within 10 degrees (Fig S4C-D) and the largest tilt that we observed was around 207 60°. However, a marked effect on cell fate determination was observed previously only when 208 the spindle angle was close to 90° (Lee et al, 2006; Cabernard and Doe, 2009) . In this case  209   spindles oriented at around 90° led to both the daughter cells assuming the NB identity, thus  210 increasing NB numbers per brain lobe. Consistent with this spindle orientation defect being 211 insufficient to cause major NB amplification, we did not see a significant difference in the NB 212 number per brain lobe between wild-type Mms19 P (Fig 1D). Therefore, the spindle orientation 213 defect due to lack of Mms19 does not appear to lead to differentiation problems, but impedes 214 efficient mitosis. 215 We conclude that in the absence of Mms19, spindle formation is not properly coordinated 216 with cell cycle progression and that defects in centrosome migration and in spindle assembly 217 and orientation contribute to the delay of the mitosis in Mms19 P NBs. 218 219 Mms19 is cell autonomously required to maintain normal cell numbers 220 In the experiments with the Mms19 P mutants, Mms19 was absent from all larval cells. In this 221 situation, the mitotic delay could be due to a systemic effect or due to the lack of a cell 222 autonomous activity of Mms19. To test whether Mms19 is specifically required in the mitotic 223 cells for the timely progression through mitosis, we generated Mms19 P mosaic NB clones in 224 an otherwise wild-type background. Mosaic clones were induced in NBs 24hrs after larval 225 hatching (ALH) and the expansion of these clones was analyzed by dissecting the brains in 226 mid-third instar larvae 72 hrs ALH. For this experiment, we focused on the type I NBs on the 227 ventral side. On average, 45 cells per clone were counted in control clones, but only around 228 30-35 cells in Mms19 p clones (P<0.05; Fig S2 A-C), indicating that the loss of Mms19 activity 229 hinders the establishment of normal cell numbers in the NB lineage. This observation 230 reaffirms our conclusion that absence of Mms19 delays mitosis in NBs and that this mitotic 231 delay probably causes more Mms19 P NBs to linger in M-phase ( Fig S1E). 232

234
Mms19 is required to form spindles of a normal length and density 235 Mms19 P spindles are generally shorter than the wild-type ones (Fig 3A-E). In normal NB 236 spindles, the centrosomes were anchored close to the cell cortex, with spindle MTs 237 emanating from them and extending to the chromosomes (Fig 3A, A'). But in around a quarter 238 of the mutant cells, even though the chromosomes were aligned at the metaphase plate, the 239 centrosomes were connected to the cell cortex and were only a short distance away from the 240 metaphase plate (Fig 3B, B'). In order to quantify this defect, we measured the length of the 241 spindles across all genotypes and found that the spindles in Mms19 P and da>CAK, Mms19 P 242 were significantly shorter than the wild-type control spindles ( Fig 3C). As the NBs were also 243 considerably smaller in Mms19 P and da>CAK, Mms19 P brains (Fig 3D), we additionally 244 displayed the spindle length relative to the cell diameter ( Fig 3E). A ratio closer to 1 indicates 245 that the centrosomes were anchored close to the cell cortex, as in a healthy spindle. On the 246 other hand, if the ratio was equal to or lower than 0.6, the spindle was defined as a 'short 247 spindle' because the centrosomes were further inside the cell. We found a significant 248 reduction in this ratio in Mms19 P and da>CAK, Mms19 P NB spindles (Fig 3E). Mms19 P NBs 249 contained around 20% short spindles while this number went up to 30% for da>CAK, Mms19 P 250 ( Fig 3F). This data indicates that the short spindle phenotype was not rescued, but actually 251 worsened upon overexpression of CAK. On the other hand, expression of Mms19::eGFP in the 252 Mms19 P background rescued the short spindle phenotype and only 3% of the NBs displayed 253 it. 254 To determine the effect of Mms19 on MT formation and stability, we measured astral as well 255 as inner spindle MT density. For astral MT quantification, we used the method described by 256 Yang and co-workers (Yang et al, 2014; Fig 3G) and found a significant reduction in the 257 Mms19 P NBs as compared to wild type (Fig 3H-I). This phenotype appeared to be slightly 258 rescued by CAK complex overexpression and was fully rescued by expressing Mms19::eGFP 259 in the Mms19 P background. Reduced astral MT stability could be linked to the spindle 260 positioning and orientation defects described previously (Fig 2B, 2E, S4B-D) as astral MTs 261 were shown to contact the cell cortex and regulate spindle positioning (Pearson et al, 2004). 262 Mms19 is thus necessary for the formation of fully assembled spindles and astral MTs. 263

265
Mms19 assists MT polymerization in vivo 266 To assess whether the short spindles could result from a defect in MT growth in vivo, we 267 studied NBs expressing EB1::GFP. Live imaging of EB1::GFP revealed the path of elongation of 268 single MTs, akin to that of a comet, and the movement speed of the GFP signal reflects the 269 growth speed of the MT plus ends. Wild-type and Mms19 P larval brains expressing EB1::GFP 270 were dissected and live imaging was performed. The speed of the EB1::GFP particles was then 271 tracked manually using ImageJ. The measurements indicated that the spindle MTs of wild-272 type NBs polymerized at a rate that is significantly higher than the rate observed in Mms19 P 273 NBs ( Fig 4A, A', WT-mov 04, Mms19 P -mov 05). To assess whether the MT assembly function 274 of Mms19 is restricted to the mitotic state, we also tested whether the absence of Mms19 275 also affects MT polymerization in the post mitotic glia cells. Measuring EB1 comet speeds in 276 glia cells showed a decrease in MT growth for Mms19 P glia as compared to wild-type glia (Fig  277   4B, B', WT-mov 06, Mms19 P -mov 07). These results established that Mms19 assists MT 278 polymerization and growth in vivo in mitotic NBs and in postmitotic glia. 279 Perhaps due to reduced MT growth, spindle assembly is also delayed in Mms19 P brain NBs. 280 Approximately 3 minutes after the onset of NEBD, a spindle fully assembled in wild-type NBs 281 ( Fig 4C). In contrast, in most Mms19 P NBs, 3 minutes after the onset of NEBD the 'spindles' 282 appear to be short and MT density seems much lower than in the wild type ( Fig 4D). In the 283 presented case, it took 7 minutes from the start of NEBD until a fully assembled spindle 284 became visible ( Fig 4D). 285

286
Mms19 is required for spindle re-polymerization 287 The short spindles observed in Mms19 P NBs might be caused by impaired MT polymerization. 288 To test whether Mms19 modulates MT polymerization dynamics, we utilized the in vivo MT 289 polymerization assay described by Gallaud and co-workers (Gallaud et al, 2014). With this 290 procedure, the NB spindles were completely depolymerized after incubation on ice for 30 min 291 ( Fig 5A). Wild-type NB spindles then regained their standard size and morphology within 90 292 sec after shifting them back to 25°C (Fig 5A). In Mms19 P NBs, however, the spindles failed to 293 re-polymerize to the normal shape after 90 sec. Instead, they remained abnormally short (Figs 294 5B, 5E). Interestingly, whereas the Mms19::eGFP fusion protein was able to rescue this 295 phenotype, CAK overexpression was unable to do so (Fig 5C-E). To validate this short spindle 296 phenotype, we also calculated the spindle sizes relative to the cell diameter after 90 sec 297 incubation at 25°C (Fig 5F). Even though Mms19 P cells are smaller, their normalized spindle 298 size was still significantly smaller than the wild-type one. When CAK was overexpressed in 299 Mms19 P NBs, the spindle became slightly longer than in Mms19 P , but it still remained shorter 300 than in the Mms19 P mutant rescued with the wild-type Mms19::eGFP transgene (Fig 5F). This 301 result shows that Mms19 has an important role in promoting spindle polymerization and 302 mainly through a CAK independent process. 303 Mms19 impedes TACC function. We therefore stained wild-type and Mms19 P NBs for TACC 313 and observed a strong signal at the wild-type centrosomes ( Fig 6A). On the other hand, in 314 >50% of Mms19 P NBs, TACC did not show any enrichment on the centrosomes (Fig 6B, D). 315 Similar results were also obtained with its interactor Msps ( Fig S6). As TACC acts downstream 316 of Aurora A kinase, and Aurora itself acts downstream of Cdk1 (Van Horn et al, 2010), this 317 defect might be caused by insufficient CAK activity in Mms19 mutants (Nag et al, 2018). We 318 therefore tested this hypothesis by over-expressing the three CAK components in the Mms19 P 319 background (da>CAK, Mms19 P ). Indeed, upon CAK overexpression, the fraction of spindles 320 displaying centrosomal TACC almost reached wild-type levels (Fig 6C, D). To quantify the 321 enrichment of TACC on centrosomes, we compared the fluorescent intensity of centrosomal 322 TACC to the TACC fluorescent intensity on the spindle. This ratio decreased for Mms19 P NBs, 323 but was rescued in da>CAK, Mms19 P NBs ( Fig 6E). It was also reported that Aurora A loss-of-324 function can cause centrosome fragmentation (Giet et al, 2002). Co-staining Mms19 P NBs with 325 antibodies against the centrosomal protein g-Tubulin showed that even in cases with 326 apparent TACC mis-localization, centrosomal g-Tubulin was still present, indicating that TACC 327 mislocalization is not caused by centrosome fragmentation. These findings indicate that 328 centrosomal localization of TACC and its stimulation of astral and spindle MT stability or 329 growth is at least partially dependent on Mms19 activity. 330 331

Mms19 binds to MTs and stimulates MT assembly 332
To explore the possibility that Mms19 also functions through different activities, we prepared 333 protein extracts from flies expressing Mms19::eGFP driven by its endogenous promoter (Nag 334 et al, 2018), subjected them to immunoprecipitations and analyzed co-purifying proteins by 335 Mass spectrometry (Supplementary Table S1). Adult wild-type flies and Imp::eGFP expressing 336 flies, respectively, served as controls to exclude any non-specific binding to beads and GFP, 337 respectively. Amongst the proteins exclusively bound to Mms19::eGFP were the CIA proteins 338 Mip18, Ciao1 and Ant2, which form a complex with Mms19 to mediate Fe-S cluster delivery 339 (Gari et al, 2012; Stehling et al, 2012). The fact that we recovered these proteins efficiently, 340 indicated that the purification was efficient. Because Mms19 functions on microtubules and 341 our second control, IMP::eGFP, is also involved in MT dependent processes, we additionally 342 inspected the data for tubulin and MT binding proteins that are enriched by the Mms19::eGFP 343 compared to the wild-type control without eGFP tag (Supplementary Table S2). This 344 comparison revealed a clear enrichment of tubulin and several Microtubule Associated 345 Proteins (MAPs). Some of the associated proteins were not present at all in the wild-type 346 control, some were present, but at lower levels compared to the Mms19::eGFP and 347 IMP::eGFP fractions. Others were exclusively found in the Mms19::eGFP fraction. These 348 results therefore suggested that Mms19 might directly or indirectly bind to MTs. 349 To validate the interaction of Mms19 with Tubulin, we next tested whether Mms19::5xHis 350 purified from E. coli and a/b-tubulin dimers interact in vitro. Mms19::5xHis was incubated at 351 an equimolar ratio with purified porcine brain a/b-tubulin and then bound to the Ni-NTA 352 resin. All incubations and washing steps were carried out at 4ᵒC. Copurifying proteins where 353 then assessed by western blotting. A band corresponding to a-tubulin was observed when 354 tubulin was incubated with Mms19::5xHis ( Fig 7A, B), but tubulin alone did not bind to the 355 resin, pointing to a direct interaction between tubulin and Mms19::5xHis. 356 To establish whether Mms19 stimulates MT polymerization, assembly or stability, we 357 performed a spectrophotometric assay in which tubulin was incubated with Mms19::5xHis at 358 a 40:1 ratio (Mirigian et al, 2013). The absorbance of this mixture was recorded at 340nm 359 over 40 min. The increase in absorbance of the sample is a measure for the polymerization of 360 MTs. The positive control, a fraction of known MT elongators (Microtubule Associated Protein 361 Rich Fraction (MAPF) produced a similar maximal absorbance (Fig 7C), but this was not seen 362 when either only Mms19 or only tubulin were added ( Fig 7C). Furthermore, to distinguish MT 363 polymerization from aggregation, we incubated this mixture after the polymerization at 4°C. 364 If tubulin indeed polymerizes into MTs, this incubation in the cold should de-polymerize the 365 MTs. On the other hand, if the increasing absorbance was due to aggregation of either tubulin 366 or Mms19, the absorbance due to aggregates would not decrease after cold treatment. 367 Because the absorbance of the sample sharply decreased in the cold, we conclude that 368 Mms19 indeed promoted MT polymerization ( recruitment to the centrosomes, and we found that not only the centrosomal localization of 422 TACC (Fig 6), but also the one of Msps depends on Mms19 and elevated CAK activity (Fig S6). 423 The activity of Mms19 through this pathway, which involves the activation of the M-phase 424 regulator Cdk1, therefore contributes significantly to the coordination of cell cycle 425 progression and spindle formation, a process that is affected in the absence of Mms19 CAK overexpression rescued TACC localization, but it could not fully rescue the short 436 spindle/spindle assembly and the microcephaly defects (Figures 1, 3, 5). Furthermore, even if 437 we did not test this directly, we can expect the glial cell MT growth defects ( Fig 4B) and the 438 neurite outgrowth defect (Fig S8) to be also independent of the CAK activity because this 439 kinase is mostly active during the mitotic phase of the cell cycle. We found that Mms19 440 interacts directly with tubulin and MTs. This is interesting because previous work had also 441 suggested an interaction with the mitotic spindle (Ito et al, 2010, van Wietmarschen et al,  442 2012, Nag et al, 2018). In this study, we report for the first time a direct interaction between 443 Mms19 and tubulin, and we linked this interaction to an activity of Mms19 in stimulating MT 444 assembly (Fig 7). The EM data indicates that Mms19 might contribute to the assembly of MTs 445 by directly binding to them, bundling them and regulating thereby the stability of MTs. 446 Interestingly, such a MT bundling activity seems also crucial for neurite outgrowth, because 447 the initial stage of this process involves bundling of MTs (Miller and Suter, 2018). This would 448 explain why this process is delayed or blocked in most Mms19 P neurons (Fig S8). 449 Recently, another connection between Fe-S cluster proteins and mitosis was described to 450 involve Kif4a in HEK293 cells (Ben-Shimon et al, 2018). However, its Drosophila homologue, 451 Klp3a, was not amongst the proteins co-purifying with Mms19::eGFP in extracts from adult 452 flies (Table S1) and the Klp3a phenotypes described by (Williams et al, 1995) Fig S7). This regulatory mechanism also links MMS19 to certain cancers, especially lung 499 squamous cell carcinomas, where MAGE-F1 was found to be upregulated (Weon et al, 2018). The following  Cy5-AACACAACACAACACAACACAACACAACACAACAC that binds to a specific region on the 2 nd 719 chromosome (Dernburg, 2011) was then added to the Hybridization buffer (20% dextran 720 sulfate, 2xSSCT, 50% Formamide) and this solution was incubated with the brains in a PCR 721 tube. The probes were then denatured at 92°C for 3 min and then allowed to anneal with the 722 chromosomal DNA overnight at 37°C. The sample was then washed thrice at 37°C with the 723 following solutions for 20min each: 2xSSCT/50% Formamide, 2xSSCT/40% Formamide, 724 2xSSCT/20% Formamide. After two more washes with 2xSSCT, the sample was stained with 725 Hoechst, mounted using Aqua-poly mount, and imaged on a Leica SP8 confocal microscope 726 with a 63x objective.  particles analyzed from each brain (C) In a WT NB, a spindle is seen to be fully assembled 965 approximately 3 minutes after NEBD. (D) On the other hand, in most Mms19 P NBs, spindle 966 assembly seemed to be delayed.